1
Closing the gap in the capacity of random wireless networks
Massimo Franceschetti2 , Olivier Dousse 3 , David Tse2 , Patrick Tiran3
Abstract — We consider the problem of how
throughput in a wireless network with randomly lo-
cated nodes scales as the number of users grows. Fol-
lowing the physical model of Gupta and Kumar, we
show that randomly scattered nodes can achieve the
√
optimal 1/ n per-node transmission rate of arbitrarily
located nodes. This contrasts with √ previous achiev-
able results suggesting that a 1/ n log n reduced rate
is the price to pay for the additional randomness intro-
duced into the system. Our results rely on percolation
theory arguments.
I. Introduction
A completely wireless network consists of n nodes that com-
municate over a common wireless channel. A natural question
that arises in such systems is how the throughput scales with
the number n. This question was first addressed in [1]. These
authors show that under the assumption that packets can be
relayed from node to node and interference is simply regarded
as noise at each hop, and if each node transmits to a destina-
tion non-vanishingly far away, the per-node throughput goes
√
to zero at least as 1/ n bit per second.
For random node locations and traffic patterns, they
√
show that a per-node throughput of rate 1/ n log n can
be achieved. Our contribution is to show that a per-node
√
throughput rate of 1/ n can also be achieved by networks of
randomly located nodes, thus closing the gap between upper
and lower bounds on the throughput capacity of such net-
works. Our proposed protocol also achieves the optimal aver-
age delay required for a packet to reach its destination.
II. Main result
We consider a random network where nodes are distributed in
a square of unit area according to a Poisson point process of
intensity n. We let the distance between two nodes i and j be
denoted by dij . We pick uniformly at random a matching of
source-destination pairs, so that each node is the destination of
exactly one source. Each node i can transmit with any power
Pi less than the maximum power P0 , and node j receives the
transmitted signal with power Pi L(i, j), where L(·) indicates
the path loss between i and j. We assume L(i, j) = d−α −γdij
ij e
for α > 0, γ > 0, or α > 2, γ = 0. We finally assume that two
nodes can establish a wireless link of unit capacity if the signal
to interference plus noise ratio at the receiver exceeds some
threshold t. That is
P L(i, j)
SIN R = !i > t,
N0 + k"=i Pk L(k, j)
1 A complete version of this paper can be found at
www.paradise.caltech.edu/∼massimo/homepage/academic.html
2 Dept. of Electrical Engineering and Computer Sciences, Univ.
of California at Berkeley, Berkeley, CA 94720-1770. e-mail:
{massimof, dtse}@EECS.berkeley.edu
3 School of Computer and Communication Sciences EPFL
CH-1015, Lausanne, Switzerland. e-mail: {Patrick.Thiran,
olivier.dousse}@epfl.ch
where N0 is the ambient noise power at the receiver. We
say that in our stochastic model an event occurs with high
probability (w.h.p.) if its probability tends to one as n → ∞.
The per-node throughput capacity of the network is defined
as the number of bits per second that every node can transmit
w.h.p. to its destination.
Our result is the following.
Theorem 1 The per-node throughput capacity of the network
defined above is
√
C(n) = Θ(1/ n) bit/sec. (1)
III. Percolation theory
To prove our result we establish a connection between perco-
lation theory [2] and wireless networks that can be potentially
exploited to solve other information and network theoretic
questions.
Let us consider the following example. Water is poured on
one side of a large (ideally infinite) porous stone. What is the
probability that the water finds a path to the opposite side?
By modeling the stone as a square grid in which each edge
can be open and hence traversed by water with probability
p, and closed otherwise, independently of all other edges, one
can show that for p > 1/2 water percolates trough the stone
with probability one. One can then ask at what rate the water
percolates and how it depends on p. In other words, how rich
in disjoint paths is the connected component of open edges?
We relate this latter question to the information flow in our
random wireless network. The open grid edges correspond to
the presence of nodes in certain locations of the plane, and the
open percolating paths represent the wireless backbone that is
used to multi-hop packets across the network. We want to op-
erate at p > 1/2, above the percolation threshold, so that we
can guarantee the existence of many paths that traverse the
network, but also have p < 1, so to avoid overcrowding and ex-
cessive interference. We show that controlling the parameter
p corresponds in a wireless network to scaling the transmission
power at a given rate as n → ∞. The scaling adopted in [1]
corresponds to p = 1, and to have a fully connected backbone
containing paths that cross the network area. However, this
wireless backbone covers all the nodes, generating excessive
,
interference. By scaling the power at a higher rate, this back-
bone does not form at all. Percolation theory ensures that a
different kind of backbone forms in the transition region be-
tween this two extreme scalings. This does not cover all the
nodes, nevertheless it is sufficiently rich in crossing paths so
that it can transport all the traffic in the network. By oper-
ating the network in this transition region between order and
disorder, we are able to prove our tight bound.
References
[1] P. Gupta, P. R. Kumar. “The capacity of wireless networks.”
IEEE Transactions on Information Theory, 46(2), 388-404,
2000.
[2] G. Grimmett. “Percolation.” Second edition, Springer Verlag,
1999.